"""
HTML Converter for Rich Console Output
This module converts Rich console output to clean HTML suitable for PDF generation.
"""
from rich.console import Console
from rich.text import Text
import re
from typing import Dict, Any, Optional
class RichToHTMLConverter:
"""Converts Rich console output to clean HTML for PDF generation."""
def __init__(self):
self.console = Console(record=True, width=120)
def rich_to_html(self, console_output: str) -> str:
"""
Convert Rich console output to clean HTML.
Args:
console_output: Raw console output with Rich formatting
Returns:
Clean HTML string suitable for PDF generation
"""
# Create a console and export to HTML
html_content = self.console.export_html(
inline_styles=True,
code_format=None
)
# Clean and optimize for PDF
return self.clean_html_for_pdf(html_content)
def clean_html_for_pdf(self, html_content: str) -> str:
"""
Clean HTML content for better PDF rendering.
Args:
html_content: Raw HTML content from Rich export
Returns:
Cleaned HTML content
"""
# Remove background colors that don't work well in PDF
html_content = re.sub(r'background-color:\s*#[0-9a-fA-F]{6};?', '', html_content)
# Ensure good contrast for text
html_content = re.sub(r'color:\s*#[0-9a-fA-F]{6};?', 'color: #333333;', html_content)
# Remove excessive margins and padding
html_content = re.sub(r'margin:\s*\d+px;?', 'margin: 5px;', html_content)
html_content = re.sub(r'padding:\s*\d+px;?', 'padding: 3px;', html_content)
return html_content
def apply_pdf_styles(self, html_content: str) -> str:
"""
Apply PDF-specific styles to HTML content.
Args:
html_content: HTML content to style
Returns:
HTML content with PDF-optimized styles
"""
pdf_styles = """
"""
# Insert styles into HTML head
if '' in html_content:
html_content = html_content.replace('', f'{pdf_styles}')
else:
html_content = f'{pdf_styles}{html_content}'
return html_content